CVE-2022-48998

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> powerpc/bpf/32: Fix Oops on tail call tests<br /> <br /> test_bpf tail call tests end up as:<br /> <br /> test_bpf: #0 Tail call leaf jited:1 85 PASS<br /> test_bpf: #1 Tail call 2 jited:1 111 PASS<br /> test_bpf: #2 Tail call 3 jited:1 145 PASS<br /> test_bpf: #3 Tail call 4 jited:1 170 PASS<br /> test_bpf: #4 Tail call load/store leaf jited:1 190 PASS<br /> test_bpf: #5 Tail call load/store jited:1<br /> BUG: Unable to handle kernel data access on write at 0xf1b4e000<br /> Faulting instruction address: 0xbe86b710<br /> Oops: Kernel access of bad area, sig: 11 [#1]<br /> BE PAGE_SIZE=4K MMU=Hash PowerMac<br /> Modules linked in: test_bpf(+)<br /> CPU: 0 PID: 97 Comm: insmod Not tainted 6.1.0-rc4+ #195<br /> Hardware name: PowerMac3,1 750CL 0x87210 PowerMac<br /> NIP: be86b710 LR: be857e88 CTR: be86b704<br /> REGS: f1b4df20 TRAP: 0300 Not tainted (6.1.0-rc4+)<br /> MSR: 00009032 CR: 28008242 XER: 00000000<br /> DAR: f1b4e000 DSISR: 42000000<br /> GPR00: 00000001 f1b4dfe0 c11d2280 00000000 00000000 00000000 00000002 00000000<br /> GPR08: f1b4e000 be86b704 f1b4e000 00000000 00000000 100d816a f2440000 fe73baa8<br /> GPR16: f2458000 00000000 c1941ae4 f1fe2248 00000045 c0de0000 f2458030 00000000<br /> GPR24: 000003e8 0000000f f2458000 f1b4dc90 3e584b46 00000000 f24466a0 c1941a00<br /> NIP [be86b710] 0xbe86b710<br /> LR [be857e88] __run_one+0xec/0x264 [test_bpf]<br /> Call Trace:<br /> [f1b4dfe0] [00000002] 0x2 (unreliable)<br /> Instruction dump:<br /> XXXXXXXX XXXXXXXX XXXXXXXX XXXXXXXX XXXXXXXX XXXXXXXX XXXXXXXX XXXXXXXX<br /> XXXXXXXX XXXXXXXX XXXXXXXX XXXXXXXX XXXXXXXX XXXXXXXX XXXXXXXX XXXXXXXX<br /> ---[ end trace 0000000000000000 ]---<br /> <br /> This is a tentative to write above the stack. The problem is encoutered<br /> with tests added by commit 38608ee7b690 ("bpf, tests: Add load store<br /> test case for tail call")<br /> <br /> This happens because tail call is done to a BPF prog with a different<br /> stack_depth. At the time being, the stack is kept as is when the caller<br /> tail calls its callee. But at exit, the callee restores the stack based<br /> on its own properties. Therefore here, at each run, r1 is erroneously<br /> increased by 32 - 16 = 16 bytes.<br /> <br /> This was done that way in order to pass the tail call count from caller<br /> to callee through the stack. As powerpc32 doesn&amp;#39;t have a red zone in<br /> the stack, it was necessary the maintain the stack as is for the tail<br /> call. But it was not anticipated that the BPF frame size could be<br /> different.<br /> <br /> Let&amp;#39;s take a new approach. Use register r4 to carry the tail call count<br /> during the tail call, and save it into the stack at function entry if<br /> required. This means the input parameter must be in r3, which is more<br /> correct as it is a 32 bits parameter, then tail call better match with<br /> normal BPF function entry, the down side being that we move that input<br /> parameter back and forth between r3 and r4. That can be optimised later.<br /> <br /> Doing that also has the advantage of maximising the common parts between<br /> tail calls and a normal function exit.<br /> <br /> With the fix, tail call tests are now successfull:<br /> <br /> test_bpf: #0 Tail call leaf jited:1 53 PASS<br /> test_bpf: #1 Tail call 2 jited:1 115 PASS<br /> test_bpf: #2 Tail call 3 jited:1 154 PASS<br /> test_bpf: #3 Tail call 4 jited:1 165 PASS<br /> test_bpf: #4 Tail call load/store leaf jited:1 101 PASS<br /> test_bpf: #5 Tail call load/store jited:1 141 PASS<br /> test_bpf: #6 Tail call error path, max count reached jited:1 994 PASS<br /> test_bpf: #7 Tail call count preserved across function calls jited:1 140975 PASS<br /> test_bpf: #8 Tail call error path, NULL target jited:1 110 PASS<br /> test_bpf: #9 Tail call error path, index out of range jited:1 69 PASS<br /> test_bpf: test_tail_calls: Summary: 10 PASSED, 0 FAILED, [10/10 JIT&amp;#39;ed]